Gemini 4 Argon vs Opus 5.5 vs DeepSeek V4.1: cost per intelligence point
A drill-race visualization of the Artificial Analysis Intelligence Index: Opus 5.5 scores 58 at $5.98 per task, Gemini 4 Argon 53 at $1.99, and DeepSeek V4.1 Flash 39 at $0.27.
Claude Opus 5.5 vs Gemini 4 Argon vs DeepSeek V4.1 Flash same benchmark. one model costs 22× more per task. I turned the Artificial Analysis Intelligence Index into a drilling race. Depth = intelligence. Every coin fed to the drill = 5¢ of compute. Opus 5.5 → 58 points, $5.98 per task Gemini 4 Argon → 53 points, $1.99 per task DeepSeek V4.1 Flash → 39 points, $0.27 per task what the race shows: → Opus is the smartest, but its last 5 points over Gemini cost 3× more → Gemini is the value sweet spot at ~27 points per dollar → DeepSeek gets 67% of Opus's score for 4.5% of the price the "best model" depends on what one extra point is worth to you. I'm racing more matchups next: speed, coding, agents. the next one drops soon. would you pay 22× more for 19 more points
This post quotes another post: x.com/yurshevv/status/2111493684189368536